Notes in-line
--
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/seminar.html
Optimising Oracle Seminar - schedule updated July 27th
"Ganesh Raja" <ganesh.raja_at_nospam.mantas.com> wrote in message
news:4134addf_1_at_mk-nntp-2.news.uk.tiscali.com...
> Vince Laurent wrote:
>
> > I have an Oracle 9.2.0.4 database. I changed a few parameters and am
> > not getting HORRIBLE db_buffer hit ratios.
> >
> > Was:
> > db_block_buffers = 250000
> > db_block_size = 8192
> > db_cache_size = 70000000
> >
> > is:
> > ##db_block_buffers = 250000
> > db_block_size = 8192
> > db_cache_size = 80000000
> >
> > I thought since going to 9i I didn't need db_block_buffers.
> > Ideas?
I suspect that you didn't have both the db_block_buffers and
the db_cache_size set at the same time, as you would have
got an error ORA-00381: cannot use new and old parameters
But your buffer size with the old settings would be ca. 2GB
(250,000 x 8K).
You buffer size with the new setting would be ca. 80M,
or about 10,000 blocks - a reduction by a factor of 25.
With a change in buffer space that extreme, it should not
surprise even the most 'BCHR is bad' fanatic that your
overall requirement for physical I/O requests should go
up, and your hit ratio drop.
> >
> > Server is HP-UX 11i
> > Thanks!
> > Vince
> > "I am mortified to be told that, in the United States of America, the
sale of a book can become a subject of inquiry, and of criminal inquiry
too."
> > - Thomas Jefferson
>
> Why do u want to follow Hit Ratios.. To Get a 100% or Name the Percent
> Hit Ratio pls visit oracledba.co.uk
>
Note that the poster is not complaining specifically about
a particular value for the BCHR - he is complaining that
he sees a significant change in the BCHR. Although most
people now realise that any 'specific' value is of itself of
little value, it is important to remember that a dramatic
change in the ratio is often indicative of a developing
(or developed) problem.
Received on Tue Aug 31 2004 - 17:14:08 CDT
